Mitochondrial Fusion

Fusion between closely apposed mitochondria is a complex process that involves the fusing of two membranes and sharing of mitochondrial contents and so requires the input of several proteins. The first of these, fuzzy onions (fzo), was discovered in Drosphila Melongaster and was found to be essential for spermatogenesis in the fruitfly as this process requires the fusion and elongation of mitochondria in postmeiotic spermatids. Fzomutant males are defective in this developmentally regulated mitochondrial fusion and are sterile (Hales and Fuller, 1997). Subsequent research led to the discovery of its yeast (Fzo1) and mammalian homologs (Mfn1 and Mfn2). These 2 large GTPase proteins were found to have 77% similarity to each other with highest homology in the GTPase domain and least conserved in C terminal region. They were also found to have overlapping but not completely redundant functions in MOM fusion but knockout

of either leads to prenatal death (Santel and Fuller, 2001, Chen et al., 2003). Both are anchored to the MOM with an N-terminal GTPase domain while the C-terminal domain is a coiled-coil exposed to the cytosol and MOM fusion is GTPase dependent. Both proteins form homo- or hetero-protein complexes and this protein oligomerization is mediated by a heptad repeat region (HR2) (Koshiba et al., 2004). These two proteins have varying expression patterns in mammalian tissue, likely relating to the rate of mitochondrial fusion in particular tissues or potential functions independent of fusion. Mfn2 is highly expressed in cardiac tissue and skeletal muscle in accordance with its role in Ca2+ signaling during mitochondrial-ER tethering, while Mfn1 has been identified as enhancing virus- responsive reporter activity suggesting a potential association with increased susceptibility to certain pathogen infections (de Brito and Scorrano, 2008) (Yoneyama et al., 2004).

The mechanism of fusion in mitochondria is still poorly understood and research is ongoing to determine the exact the method but it is believed to involve GTP hydrolysis. The best-studied model of membrane fusion is in synaptic vesicle fusion with the neuronal outer membrane that utilizes a SNARE (soluble N- ethylmaleimide-sensitive factor attachment protein receptor) complex. Firstly, the vesicle becomes ‘tethered’ to the target membrane although a gap remains between them. This step is mediated by a group of GTPases on the vesicle membrane known as Rab guanosine triphosphatases. Next, SNARE proteins on both membranes form a complex leading to the membranes attaining a closer proximity in a state termed ‘docking’. Finally, the two bilayers fuse owing to the close proximity brought about by the SNARE complex formation. Interestingly, the defining feature of SNARE proteins is an extended coiled-coil stretch as is seen in the C-terminal of the mitofusins (Jahn and Fasshauer, 2012). It is believed that Mfn1 and 2 act as SNARE proteins and research published by Koshiba et al. (2004) provided us with a detailed mechanism of how MOM fusion is carried out. The trans homo-and hetero-oligomeric complexes formed by Mfn1/2 tether apposing membranes together with a distance of around 95Å separating them. This ‘tethering’ is believed to be mediated by the HR2 regions of either protein as

when they are isolated from the total protein, homo- and heterotypic complexes are formed. Mfn1 appears to be particularly important in this tethering stage as they showed that cells expressing a mutant form of Mfn1 without any GTPase activity showed a mitochondrial phenotype in which they were aggregated and trapped in a “tethered” state. Coupled to this, exogenous expression of wild type Mfn1 can rescue cells expressing Mfn2 disease mutants but expression of wild type Mfn2 cannot (Detmer and Chan, 2007). Further evidence for an interaction between both mitofusins was provided when it was shown that a heterotypic Mfn1-Mfn2 trans complex had greater efficacy in fusion events than homotypic Mfn1 or Mfn2 complexes (Hoppins et al., 2011). Furthermore, binding of HR1 and HR2 in homotypic Mfn2 complexes have been shown to be inhibitory to MOM fusion (Huang et al., 2011). A negative regulator of Mfn1/2 has also been identified in mitofusin binding protein (MIB). MIB is a member of the medium chain dehydrogenase/reducatase protein superfamily and overexpression in HeLa cells results in mitochondrial fragmentation while gene silencing results in an elongated mitochondrial morphology (Eura et al., 2006). Despite these advances in modeling and comparisons to SNARE vesicle fusion, the mechanistic details of fusion that follow GTP hydrolysis have yet to be elucidated.

The process of fusion between two mitochondria is quick and is completed in less than 2 minutes, however it can occur in the absence of any nuclear instruction or de novo translation which suggests it’s an autonomous process occurring in response to cellular cues (Schauss et al., 2010). It should be noted that despite the similarities, SNARE proteins range in size from 80 to 100 amino acids while Mfn1 and 2 are 741 and 757 amino acids long respectively indicating their role as multifunctional receptors with additional roles independent of fusion. The proposed major function of mitochondrial fusion is the sharing of mitochondrial contents to maintain heterogeneity yet this appears to be an insignificant function for a process that is vital for organismal survival (Chen et al., 2003, Davies et al., 2007). Instead it seems more likely that the key role of mitochondrial fusion is to marry the bioenergetic state of the cell with mitochondrial function, and some of the ways in which it does this is through sharing of mtDNA content and alteration

of mitochondrial morphology. This idea is supported by the fact that fusion can be inhibited or activated by the cytosolic milieu including phosphorylation events, oxidized glutathione or other post translational modification including O- GlcNAcylation (Pyakurel et al., 2015, Shutt et al., 2012). The other protein involved in mitochondrial fusion, Opa1, supports this hypothesis as it acts as a cellular stress sensor in order to modulate mitochondrial function.

Opa1, or Optic Atrophy type 1, is a large GTPase required for inner mitochondrial membrane fusion, so called because mutations in the gene are the most common cause of this dominantly inherited optic neuropathy that leads to progressive loss in visual acuity and eventually blindness (Delettre et al., 2000). It is a dynamin- like protein with a sequence homology to the yeast gene Mgm1, responsible for maintaining mtDNA and inner membrane structures in yeast while also interacting with Fzo1 to regulate mitochondrial fusion. Unlike Mfn1/2, Opa1 is only required on one MIM for mitochondrial fusion to occur (Song et al., 2009a). Exogenous expression of OPA1 in mammalian cell lines results in an altered mitochondrial morphology while knockdown using siRNA results in fragmented mitochondria proving it’s involvement in MIM fusion (Ishihara et al., 2006). Opa1 is targeted to mitochondria via a basic-rich N-terminal targeting sequence where it is anchored within the MIM facing the intra membrane space (Olichon et al., 2002). Opa1 functions are controlled by a complex pattern of alternative splicing and proteolysis. There are 8 different isoforms which all have the mitochondrial leader sequence mentioned above. This is cleaved by the mitochondrial processing peptidase (MPP) upon import into the mitochondria (Head et al., 2009). From here, Opa1 processing follows two distinct directions. If the isoform contains alternatively spliced exons 4b or 5b, they are constitutively cleaved by the intra- membrane space AAA protease YME1L1 which generates the short form of Opa1 (S-Opa1) while the remaining isoforms are not cleaved further generating the long form (L-Opa1). This process is regulated by a group of MIM proteins known as prohibtins (Otera and Mihara, 2011, Griparic et al., 2007). The shorter isoforms are more loosely bound to the IM than the long forms that retain their hydrophobic domain and it appears that this may have a functional significance (Ishihara et al.,

2006). Loss of ΔΨm results in a significant fragmentation of mitochondria which is closely linked to a rapid conversion of L-Opa1 to S-Opa1 mediated by Oma1, a protease with multiple membrane spanning segments and a zinc binding motif (McBride and Soubannier, 2010).

It is unclear if MIM and MOM fusion occur simultaneously or sequentially or completely separately. The two processes can be separated by dissipation of the inner membrane potential with valinomycin or CCCP as shown by Malka et al. (2005) but they also demonstrated that they proceed separately in the absence of any pharmacological interference suggesting they can function separately. This was further backed up by evidence that MOM fusion can proceed in OPA1 null MEFs and that OPA1 is not required on adjacent mitochondria for MIM fusion as OPA1-null/wildtype cell hybrids exhibited mostly partial mitochondrial fusion (Song et al., 2009b). Finally MIM fusion appears to be Mfn-1 dependent (Guillery et al., 2008) so we can conclude that MOM fusion can proceed without MIM fusion but the latter requires the presence of Mfn-1. Increased OXPHOS activity and higher levels of ATP are associated with elongated mitochondria as this type of mitochondrion appears to have more efficient energy generation and is capable of distributing energy over long distances (Mitra et al., 2009, Skulachev, 2001). This likely occurs as increased OXPHOS activity can then stimulate inner membrane fusion while outer membrane fusion remains unaffected (Mishra et al., 2014) further highlighting the separation of the processes and the role of a metabolic sensor via it’s alternative proteolysis by Yme1L or Oma1. Opa1 has further roles in cristae remodeling and supercomplex formation indicating the importance of other roles for these fusion proteins in cellular metabolism and how closely linked mitochondrial bioenergetics and dynamics are (Cogliati et al., 2013).

Figure 1.4 Putative Mechanisms of mitochondrial fusion and fission. During fission, Dynamin related protein 1 (Drp1) is dephosphorylated by calcineurin and recruited to the outer mitochondrial membrane. Once recruited, Drp1 forms an oligomeric ring around the outer membrane with the aid of the microtubule cytoskeleton. Upon GTP hydrolysis, this ring structure constricts causing membrane scission and fission of the mitochondrion. During fusion, Mitofusins 1 and 2 (Mfn1/2) assemble in homo- and hetero oligomers and mediate fusion via apposition of mitochondria in trans. Optic Atrophy 1 (OPA1) is localized to the inner mitochondrial membrane and plays an important role in inner membrane fusion although the exact mechanism of action is currently unclear. Adapted from Dorn and Kitsis (2015).